If the study I read this morning holds up, maybe you and I and everybody else should start carrying a wallet-sized photo of ourselves. Make it one that's cute or fun or expressive. Maybe call attention to it with a big bright smiley on the front. And attach this note: IF YOU ARE READING THIS IN THE EMERGENCY ROOM AND I AM GETTING A CT SCAN, PLEASE PLEASE PLEASE GIVE MY PICTURE TO THE RADIOLOGIST.
Why? Because when faces of patients who got a diagnostic CT scan were put on the screen along with the CT images, radiologists did a more thorough job of interpreting the results. The study, presented today at the radiologists' annual scrum, hosted in Chicago by the Radiological Society of North America, was small, but its findings were clear, significant, and either intriguing or scary, depending whether you're Pollyannaish or a skeptical reporter. My reaction was alarm at what is now being missed rather than elation at the additional information that might be unearthed.
The study was devised by Yehonatan Turner, a radiology resident at Shaare Zedek Medical Center in Jerusalem who was thinking about ways to stay completely focused on every one of the scores of images he reviewed in the course of his ER duties. "I was afraid I was getting used to the routine," Turner told me. "I tried to find a way to treat each CT scan differently, and I thought about attaching the patient's face." Not all of the ER staff was pleased about the addition of another consent form and a photo to the busy activity, but he prevailed.
More than 300 patients were recruited who had come to the ER and received a CT scan to diagnose strokes, abdominal injuries, and other emergency ailments. The study team selected 96 CT images at random and parceled them out to senior radiologists to review once and then again after three months, with the same radiologists re-evaluating the same images. One group of radiologists looked at the scans with the patients' faces attached initially but not the second time. Another group saw the faces the second time but not the first. And a control group didn't see faces either time. (This group was too small to be statistically significant but provided the researchers with a rough comparison.)
The key finding was that when the patients' faces were on the screen, the radiologists came up with almost 50 percent more "incidental findings"—indications that are not part of the reason for the scan but that the radiologists are expected to look for, such as a kidney stone in a patient being checked for appendicitis. (Other media accounts of the study have cited an 80 percent difference, but that is from a preliminary version of the study.) On average, the radiologists identified 4.0 incidental findings with the faces shown and 2.7 when they weren't.
That's not all. When the radiologists dictated their reports, the ones based on scans with faces were significantly longer, averaging about 60 words, compared with about 45. Far more radiologists—50 against 29—included a summary of their findings in the face-based report. And 17 added recommendations, compared with four for reports on faceless patients.
Summaries and recommendations are not medically required, says Turner; it is more a matter of physician preference. "Sometimes we write one, sometimes we don't," he says. "If there is a prominent finding, we write one." But inclusion sends a message. "It is a sign that the physician was more involved with the patient, more devoted to the exam," Turner says. A survey of the radiologists at the conclusion of the study turned up such comments as "The patient's photograph prompted me to relate in more detail to the CT" and "The presence of the photograph enabled me to feel [like] more of a physician while evaluating the exam."
Turner is commendably cautious. "We have to validate these results," he says. "I think that we miss a lot of incidental findings, but this study is not the final verdict." And he points out that the vast majority of incidental findings are not life threatening.
Still, I would want to know about them. As I assume you would—wouldn't you?